Methodological suggestions which numerous researchers have suggested might explicate the discrepancy were incorporated. These include the use of a disproportionate stratified sample to facilitate focus on relatively serious self-reported delinquent behavior, separate analyses of subscales representing different domains of delinquency, and operationalization of social class by multiple indicators, including some consistent with the concept of an underclass. It was found that only offenses against persons consistently correlate negatively with measures of social class. However, analysis of self-reported contacts with police and courts suggests the presence of social class biases in the juvenile justice process and raises questions concerning the appropriateness of using disproportionate stratified samples in this context. (Author abstract)